
Building reasoning models @OpenAI. ICPC World Champion
London, England
Joined August 2010
- Tweets374
- Following351
- Followers4.4K
- Likes640
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
Borys Minaiev retweeted
Introducing Unreal Agent:
An open-source harness with state-of-the-art cost efficiency
39% cheaper than Codex+Astra on Terminal-Bench 4.0 while maintaining performance
The idea is pretty cool! Good luck!
meet @mostik_ai!
what happens when you put 12 PhDs in one room for four months? first place on the ARC-AGI leaderboard, which I can't say much about while the competition is still running. and this, which I can.
everyone's arguing about whether open models will catch up to frontier models. we think it's the wrong question. here's the one we pose: why does a frontier model have to generate your answer at all, when the only thing you need from it is the reasoning?
we do this by enabling models to communicate in latent space. through our protocol, hidden states pass straight from a frontier model into a small one running on your infrastructure -- no text between them, and neither model is fine-tuned. two models from different families, sharing reasoning, both left untouched.
how do we know it works? we tested it on a setup where a 753B model reads the problem, and a 4B edge-class model writes the answer. with this approach, we get results 80% as accurate as the frontier model, but at 20x faster performance.
we're committed to preventing frontier model lock-in and are already partnering with inference providers to accelerate open-weight adoption. we've done this between 15 of us, in four months, 12 PhDs and a Fields medalist, backed by @generalcatalyst
WIRED has the first external account of the company and the work: wired.com/story/russian-star…
full writeup, the setup, and all the numbers: mostik.ai/read-more
1
26
1,984
Borys Minaiev retweeted
Yesterday, my OpenAI collaborator and I gave a detailed talk on the Huggingface incident, our models creating "the message board", model misalignment, and more.
youtube.com/watch?v=87DyyMV0…
I hope it can answer a lot of the questions folks have, and we will release a full detailed postmortem at a later time!
Borys Minaiev retweeted
Highly recommend participating in the Midnight Code Cup contest. It's going to be a very unique experience, since the finals are 24h long and (AFAIK) it's the only onsite contest where using AI agents is not only allowed but highly encouraged.
If you've ever fantasized about participating in a contest, that's honestly your best bet. Those contests are designed to be weird and people who have done competitive programming for 10 years won't have any real advantage over you.
Some details:
- it's a team contest (3 people)
- qualifications are online (April 11, 4h online on codeforces)
- finals are in Belgrade, Serbia (July 4-5, 24h)
- organizers' write-up from previous edition: codeforces.com/blog/entry/14…
- more details: midnightcodecup.org/
Last but not least, it's unclear how many such contests we're going to have, so don't miss your chance!
Registration open now at midnightcodecup.com. See you at Midnight!
Borys Minaiev retweeted
My first blogpost is out: ikot.blog/anthropic-take-hom…
Like many of you, my feed has been dominated recently by Anthropic's Optimization team take-home.
TL;DR: They retired this "notoriously difficult" exam because Claude Opus 4.5 effectively solved it. So, they released the task, and everyone started the grind.
Yes, AI beat human candidates here, but AI-generated solutions aren't easy to follow if you aren't familiar with the domain. And they carry 0 educational value.
I didn't just want to see the solution; I wanted to understand the mechanics under the hood.
So I spent the weekend digging into the task. I started with a naive Python baseline that took 147k cycles. After three rounds of specific optimizations, I got it down to ~2,200 cycles.
That’s a 65x speedup—and likely would have passed the hiring bar half a year ago.
I’ve just published a full breakdown of the solution. I explain everything in detail and depth, but I kept it accessible. Hope you'll love the visuals!
Enjoy!
New on the Anthropic Engineering Blog: We give prospective performance engineering candidates a notoriously difficult take-home exam. It worked well—until Opus 4.5 beat it.
Here's how we designed (and redesigned) it: anthropic.com/engineering/AI…
Borys Minaiev retweeted
Sharing OpenAI's submissions that solved 12 out of 12 problems from the ICPC World Finals, under the same 5-hour constraint and judged concurrently and identically to championship submissions.
github.com/openai/openai-icp…
Replying to @MostafaRohani
2/n
We officially competed in the onsite AI track of the ICPC, with the same 5-hour time limit to solve all twelve problems, submitting to the ICPC World Finals Local Judge - judged identically and concurrently to the ICPC World Championship submissions. We received the problems in the exact same PDF form, and the reasoning system selected which answers to submit with no bespoke test-time harness whatsoever. For 11 of the 12 problems, the system’s first answer was correct. For the hardest problem, it succeeded on the 9th submission. Notably, the best human team achieved 11/12.
Borys Minaiev retweeted
1/n
I’m really excited to share that our @OpenAI reasoning system got a perfect score of 12/12 during the 2025 ICPC World Finals, the premier collegiate programming competition where top university teams from around the world solve complex algorithmic problems. This would have placed it first among all human participants. 🥇🥇
1/5 In 2015, I won the ICPC World Finals as a member of the ITMO University team. It was the only time in Finals history when a team solved all the problems before the contest ended.
4/5 Progress is very fast! A year ago, AI struggled with even easy contest problems. Now it performs better than the best human teams. If this trend continues, next year we may see real scientific discoveries made by AI.
Borys Minaiev retweeted
2/n We officially competed in the online AI track of the IOI, where we scored higher than all but 5 (of 330) human participants and placed first among AI participants. We had the same 5 hour time limit and 50 submission limit as human participants. Like the human contestants, our system competed *without* internet or RAG, and just access to a basic terminal tool.
Very nice analysis of the system we have built for solving heuristic contests!
19
6,023
Borys Minaiev retweeted
1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).
Borys Minaiev retweeted
we're competing in the @atcoder World Finals programming contest.
real nailbiter — OpenAI has been #1 for most of the contest. looked like it might be over when @FakePsyho pulled ahead, but we've just retaken the lead.
1 hour and 20 minutes to go!
【AWTF2025】
The finals begin tomorrow!
Both days will be streamed live!
Twelve top finalists will compete in each of the Heuristic and Algorithm divisions.
Don't miss the intense battles live from Shibuya!
▼ July 16 – Heuristic Division 📺 youtube.com/live/TG3ChQH61vE
Borys Minaiev retweeted
300+ teams. 68 countries. 24 hours.
Only 30 made it to the finals. Only 1 will win.
On June 7-8, 30 finalist teams will gather in Belgrade for Midnight Code Cup — a 24-hour engineering battle judged by ICPC world champions
Powered by Recraft and partners, built by world champions. Here’s what makes MCC different 🧵👇





























